1 /* $NetBSD: ext2fs_readwrite.c,v 1.24 2002/03/25 02:23:55 chs Exp $ */ 2 3 /*- 4 * Copyright (c) 1997 Manuel Bouyer. 5 * Copyright (c) 1993 6 * The Regents of the University of California. All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 * 36 * @(#)ufs_readwrite.c 8.8 (Berkeley) 8/4/94 37 * Modified for ext2fs by Manuel Bouyer. 38 */ 39 40 #include <sys/cdefs.h> 41 __KERNEL_RCSID(0, "$NetBSD: ext2fs_readwrite.c,v 1.24 2002/03/25 02:23:55 chs Exp $"); 42 43 #include <sys/param.h> 44 #include <sys/systm.h> 45 #include <sys/resourcevar.h> 46 #include <sys/kernel.h> 47 #include <sys/file.h> 48 #include <sys/stat.h> 49 #include <sys/buf.h> 50 #include <sys/proc.h> 51 #include <sys/conf.h> 52 #include <sys/mount.h> 53 #include <sys/vnode.h> 54 #include <sys/malloc.h> 55 #include <sys/signalvar.h> 56 57 #include <ufs/ufs/inode.h> 58 #include <ufs/ext2fs/ext2fs.h> 59 #include <ufs/ext2fs/ext2fs_extern.h> 60 61 62 #define doclusterread 0 /* XXX underway */ 63 #define doclusterwrite 0 64 65 /* 66 * Vnode op for reading. 67 */ 68 /* ARGSUSED */ 69 int 70 ext2fs_read(v) 71 void *v; 72 { 73 struct vop_read_args /* { 74 struct vnode *a_vp; 75 struct uio *a_uio; 76 int a_ioflag; 77 struct ucred *a_cred; 78 } */ *ap = v; 79 struct vnode *vp; 80 struct inode *ip; 81 struct uio *uio; 82 struct m_ext2fs *fs; 83 struct buf *bp; 84 void *win; 85 vsize_t bytelen; 86 ufs_daddr_t lbn, nextlbn; 87 off_t bytesinfile; 88 long size, xfersize, blkoffset; 89 int error; 90 91 vp = ap->a_vp; 92 ip = VTOI(vp); 93 uio = ap->a_uio; 94 95 #ifdef DIAGNOSTIC 96 if (uio->uio_rw != UIO_READ) 97 panic("%s: mode", "ext2fs_read"); 98 99 if (vp->v_type == VLNK) { 100 if ((int)ip->i_e2fs_size < vp->v_mount->mnt_maxsymlinklen || 101 (vp->v_mount->mnt_maxsymlinklen == 0 && 102 ip->i_e2fs_nblock == 0)) 103 panic("%s: short symlink", "ext2fs_read"); 104 } else if (vp->v_type != VREG && vp->v_type != VDIR) 105 panic("%s: type %d", "ext2fs_read", vp->v_type); 106 #endif 107 fs = ip->i_e2fs; 108 if ((u_int64_t)uio->uio_offset > 109 ((u_int64_t)0x80000000 * fs->e2fs_bsize - 1)) 110 return (EFBIG); 111 if (uio->uio_resid == 0) 112 return (0); 113 if (uio->uio_offset >= ip->i_e2fs_size) 114 return (0); 115 116 if (vp->v_type == VREG) { 117 error = 0; 118 while (uio->uio_resid > 0) { 119 bytelen = MIN(ip->i_e2fs_size - uio->uio_offset, 120 uio->uio_resid); 121 122 if (bytelen == 0) { 123 break; 124 } 125 win = ubc_alloc(&vp->v_uobj, uio->uio_offset, 126 &bytelen, UBC_READ); 127 error = uiomove(win, bytelen, uio); 128 ubc_release(win, 0); 129 if (error) { 130 break; 131 } 132 } 133 goto out; 134 } 135 136 for (error = 0, bp = NULL; uio->uio_resid > 0; bp = NULL) { 137 if ((bytesinfile = ip->i_e2fs_size - uio->uio_offset) <= 0) 138 break; 139 lbn = lblkno(fs, uio->uio_offset); 140 nextlbn = lbn + 1; 141 size = fs->e2fs_bsize; 142 blkoffset = blkoff(fs, uio->uio_offset); 143 xfersize = fs->e2fs_bsize - blkoffset; 144 if (uio->uio_resid < xfersize) 145 xfersize = uio->uio_resid; 146 if (bytesinfile < xfersize) 147 xfersize = bytesinfile; 148 149 if (lblktosize(fs, nextlbn) >= ip->i_e2fs_size) 150 error = bread(vp, lbn, size, NOCRED, &bp); 151 else { 152 int nextsize = fs->e2fs_bsize; 153 error = breadn(vp, lbn, 154 size, &nextlbn, &nextsize, 1, NOCRED, &bp); 155 } 156 if (error) 157 break; 158 159 /* 160 * We should only get non-zero b_resid when an I/O error 161 * has occurred, which should cause us to break above. 162 * However, if the short read did not cause an error, 163 * then we want to ensure that we do not uiomove bad 164 * or uninitialized data. 165 */ 166 size -= bp->b_resid; 167 if (size < xfersize) { 168 if (size == 0) 169 break; 170 xfersize = size; 171 } 172 error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio); 173 if (error) 174 break; 175 brelse(bp); 176 } 177 if (bp != NULL) 178 brelse(bp); 179 180 out: 181 if (!(vp->v_mount->mnt_flag & MNT_NOATIME)) { 182 ip->i_flag |= IN_ACCESS; 183 if ((ap->a_ioflag & IO_SYNC) == IO_SYNC) 184 error = VOP_UPDATE(vp, NULL, NULL, UPDATE_WAIT); 185 } 186 return (error); 187 } 188 189 /* 190 * Vnode op for writing. 191 */ 192 int 193 ext2fs_write(v) 194 void *v; 195 { 196 struct vop_write_args /* { 197 struct vnode *a_vp; 198 struct uio *a_uio; 199 int a_ioflag; 200 struct ucred *a_cred; 201 } */ *ap = v; 202 struct vnode *vp; 203 struct uio *uio; 204 struct inode *ip; 205 struct m_ext2fs *fs; 206 struct buf *bp; 207 struct proc *p; 208 ufs_daddr_t lbn; 209 off_t osize; 210 int blkoffset, error, flags, ioflag, resid, xfersize; 211 vsize_t bytelen; 212 void *win; 213 off_t oldoff; 214 boolean_t async; 215 216 ioflag = ap->a_ioflag; 217 uio = ap->a_uio; 218 vp = ap->a_vp; 219 ip = VTOI(vp); 220 error = 0; 221 222 #ifdef DIAGNOSTIC 223 if (uio->uio_rw != UIO_WRITE) 224 panic("%s: mode", "ext2fs_write"); 225 #endif 226 227 switch (vp->v_type) { 228 case VREG: 229 if (ioflag & IO_APPEND) 230 uio->uio_offset = ip->i_e2fs_size; 231 if ((ip->i_e2fs_flags & EXT2_APPEND) && 232 uio->uio_offset != ip->i_e2fs_size) 233 return (EPERM); 234 /* FALLTHROUGH */ 235 case VLNK: 236 break; 237 case VDIR: 238 if ((ioflag & IO_SYNC) == 0) 239 panic("%s: nonsync dir write", "ext2fs_write"); 240 break; 241 default: 242 panic("%s: type", "ext2fs_write"); 243 } 244 245 fs = ip->i_e2fs; 246 if (uio->uio_offset < 0 || 247 (u_int64_t)uio->uio_offset + uio->uio_resid > 248 ((u_int64_t)0x80000000 * fs->e2fs_bsize - 1)) 249 return (EFBIG); 250 /* 251 * Maybe this should be above the vnode op call, but so long as 252 * file servers have no limits, I don't think it matters. 253 */ 254 p = uio->uio_procp; 255 if (vp->v_type == VREG && p && 256 uio->uio_offset + uio->uio_resid > 257 p->p_rlimit[RLIMIT_FSIZE].rlim_cur) { 258 psignal(p, SIGXFSZ); 259 return (EFBIG); 260 } 261 if (uio->uio_resid == 0) 262 return (0); 263 264 async = vp->v_mount->mnt_flag & MNT_ASYNC; 265 resid = uio->uio_resid; 266 osize = ip->i_e2fs_size; 267 268 if (vp->v_type == VREG) { 269 while (uio->uio_resid > 0) { 270 oldoff = uio->uio_offset; 271 blkoffset = blkoff(fs, uio->uio_offset); 272 bytelen = MIN(fs->e2fs_bsize - blkoffset, 273 uio->uio_resid); 274 275 error = ext2fs_balloc_range(vp, uio->uio_offset, 276 bytelen, ap->a_cred, 0); 277 if (error) { 278 break; 279 } 280 win = ubc_alloc(&vp->v_uobj, uio->uio_offset, 281 &bytelen, UBC_WRITE); 282 error = uiomove(win, bytelen, uio); 283 ubc_release(win, 0); 284 if (error) { 285 break; 286 } 287 288 /* 289 * update UVM's notion of the size now that we've 290 * copied the data into the vnode's pages. 291 */ 292 293 if (vp->v_size < uio->uio_offset) { 294 uvm_vnp_setsize(vp, uio->uio_offset); 295 } 296 297 /* 298 * flush what we just wrote if necessary. 299 * XXXUBC simplistic async flushing. 300 */ 301 302 if (!async && oldoff >> 16 != uio->uio_offset >> 16) { 303 simple_lock(&vp->v_interlock); 304 error = VOP_PUTPAGES(vp, (oldoff >> 16) << 16, 305 (uio->uio_offset >> 16) << 16, PGO_CLEANIT); 306 } 307 } 308 if (error == 0 && ioflag & IO_SYNC) { 309 simple_lock(&vp->v_interlock); 310 error = VOP_PUTPAGES(vp, trunc_page(oldoff), 311 round_page(blkroundup(fs, uio->uio_offset)), 312 PGO_CLEANIT | PGO_SYNCIO); 313 } 314 315 goto out; 316 } 317 318 flags = ioflag & IO_SYNC ? B_SYNC : 0; 319 for (error = 0; uio->uio_resid > 0;) { 320 lbn = lblkno(fs, uio->uio_offset); 321 blkoffset = blkoff(fs, uio->uio_offset); 322 xfersize = MIN(fs->e2fs_bsize - blkoffset, uio->uio_resid); 323 if (xfersize < fs->e2fs_bsize) 324 flags |= B_CLRBUF; 325 else 326 flags &= ~B_CLRBUF; 327 error = ext2fs_balloc(ip, 328 lbn, blkoffset + xfersize, ap->a_cred, &bp, flags); 329 if (error) 330 break; 331 if (ip->i_e2fs_size < uio->uio_offset + xfersize) { 332 ip->i_e2fs_size = uio->uio_offset + xfersize; 333 } 334 error = uiomove((char *)bp->b_data + blkoffset, xfersize, uio); 335 336 /* 337 * update UVM's notion of the size now that we've 338 * copied the data into the vnode's pages. 339 */ 340 341 if (vp->v_size < uio->uio_offset) { 342 uvm_vnp_setsize(vp, uio->uio_offset); 343 } 344 345 if (ioflag & IO_SYNC) 346 (void)bwrite(bp); 347 else if (xfersize + blkoffset == fs->e2fs_bsize) 348 bawrite(bp); 349 else 350 bdwrite(bp); 351 if (error || xfersize == 0) 352 break; 353 } 354 355 /* 356 * If we successfully wrote any data, and we are not the superuser 357 * we clear the setuid and setgid bits as a precaution against 358 * tampering. 359 */ 360 361 out: 362 ip->i_flag |= IN_CHANGE | IN_UPDATE; 363 if (resid > uio->uio_resid && ap->a_cred && ap->a_cred->cr_uid != 0) 364 ip->i_e2fs_mode &= ~(ISUID | ISGID); 365 if (error) { 366 (void) VOP_TRUNCATE(vp, osize, ioflag & IO_SYNC, ap->a_cred, 367 uio->uio_procp); 368 uio->uio_offset -= resid - uio->uio_resid; 369 uio->uio_resid = resid; 370 } else if (resid > uio->uio_resid && (ioflag & IO_SYNC) == IO_SYNC) 371 error = VOP_UPDATE(vp, NULL, NULL, UPDATE_WAIT); 372 KASSERT(vp->v_size == ip->i_e2fs_size); 373 return (error); 374 } 375